基于选择判据和贴近度的D-S证据融合方法

D-S Evidence Combination Method Based on Selection Standard and Closeness Degree

摘要:

针对传统D-S证据理论难以融合高度冲突证据的问题,并考虑到证据正常时Dempster规则具有优越的聚焦性能,提出了一种基于选择判据和贴近度的证据融合方法。把贴近度概念引入到D-S证据合成中,通过证据的一致性度量来计算证据的权重,从而实现了冲突证据的加权融合。同时提出了证据修正的选择判据,将证据分成冲突与非冲突两类,对冲突的证据进行修正后再进行合成,而非冲突证据可直接进行合成。通过实例验证表明,所提出的方法不但保持了Dempster规则优越的信息聚焦性能,而且较好的解决了冲突证据的合成问题。

Abstract:

Aimed at the traditional D-S evidence theory is difficult to fuse the highly conflict evidence problem, and considered the Dempster rule has a superior focusing performance when evidence is normal, an improved combination algorithm based on selection standard and closeness degree is proposed. Introduce the concept of closeness to the D-S evidence combination, through the evidence of consistencySvalue to calculate its weight, so as to realize a weighted fusion of conflict evidence. SelectionSstandard of modified evidence is proposed at the same time, the combination of evidence can be divided into conflict and non-conflicting, and fusing the evidence with the improved algorithm and the traditional algorithm respectively. Through the example show that the proposed method not only have the information focus functions of the Dempster rule, but also better solved the problem of the combination of conflict evidence.